Smseow Peekaboo

WarnAudited by ClawScan on May 16, 2026.

Overview

This skill openly enables broad AI desktop control, including screenshots, clicks, typing, hotkeys, and desktop-control servers, but gives little scoping or confirmation guidance.

Install only if you intentionally want an agent to operate your desktop. Verify the external tools first, grant permissions only during a trusted session, keep sensitive apps closed, require confirmation before any click/type/hotkey action, and stop or revoke desktop-control access when finished.

Findings (4)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

High

#ASI02: Tool Misuse and Exploitation

What this means

If used carelessly, the agent could click the wrong control, type into the wrong app, expose sensitive on-screen content, or trigger account and system actions.

Why it was flagged

These commands give the agent broad ability to view the screen and operate the user's desktop, including clicking, typing, and sending hotkeys, with no stated confirmation or scope limits.

Skill content

peekaboo image --mode screen
peekaboo click --on "按钮文字"
peekaboo type --text "内容"
peekaboo hotkey cmd,c

Recommendation

Use only in a controlled session, require explicit approval before clicks/typing/hotkeys, avoid sensitive apps while active, and prefer a VM or test account for automation.

Medium

#ASI03: Identity and Privilege Abuse

What this means

Granting this permission can let the desktop-control tool act across applications, not just within one app.

Why it was flagged

The skill tells users to grant macOS accessibility-style control, which is expected for desktop automation but is a broad local privilege.

Skill content

peekaboo permissions grant
...
⚠️ macOS 需要 15+，授权辅助功能

Recommendation

Grant permissions only to a verified trusted binary, monitor its use, and revoke the permission when desktop control is no longer needed.

Medium

#ASI04: Agentic Supply Chain Vulnerabilities

What this means

A changed, compromised, or mistaken package could receive the same desktop-control permissions the user grants for this workflow.

Why it was flagged

The setup relies on unpinned third-party packages and a cloned GitHub repository for powerful desktop-control functionality.

Skill content

npx -y @steipete/peekaboo
pip install mcp-desktop-control
npx -y mcp-desktop-pro serve
git clone https://github.com/bhyoo/kwin-mcp
python kwin-mcp.py

Recommendation

Verify the package/repository source, pin versions or commit hashes where possible, and review the tool before granting accessibility or desktop-control access.

Medium

#ASI07: Insecure Inter-Agent Communication

What this means

Screen contents and desktop actions may flow through an MCP tool channel to the controlling agent, so misconfiguration could expose private on-screen information or control capability.

Why it was flagged

The Linux path uses MCP desktop-control tooling, but the artifact does not describe connection boundaries, client authorization, or how screenshot/control access is limited.

Skill content

Linux (桌面 MCP)
...
mcp-desktop-control
...
npx -y mcp-desktop-pro serve

Recommendation

Run MCP desktop-control tools only locally or on a trusted channel, limit which agents can connect, and stop the server when not actively using it.